iPad Pro M4 Geekbench scores wow…

[ipaddaro]

If workloads can use all cores effectively, then the efficiency cores contribute a significant boost. And the M4 has 50% more efficiency cores than M3. ie. M4 loses one performance core, but gains two efficiency cores. Benchmarks like this are designed to utilize all cores effectively.

However, these benchmarks may not always translate to real world applications, because some applications may not benefit as much as others from so many extra cores.

thanks! in your opinion, apart from real usage, there are other benchmark apps that could provide a more complete test?

 

[EugW]

thanks! in your opinion, apart from real usage, there are other benchmark apps that could provide a more complete test?

Just wait until next week. Once the units are people's hands, you'll get to check out a bazillion different benchmark comparisons.

 

[ipaddaro]

Just wait until next week. Once the units are people's hands, you'll get to check out a bazillion different benchmark comparisons.

at that time I’ll also have mine.. I’m just curious how to test it properly when arrives

 

[Lounge vibes 05]

Curiously that benchmark has the version running as iOS 17.6…
The other benchmark from the 10 core was definitely on 17.5

 

[Maven1975]

I’d like to see sustained scores. I find it hard to believe it will be able to do this consistently.

Since the ‘‘improved“ heat sink is now behind the Apple logo, hopefully someone can develop a stand with built-in an AIO water pump.

 

[EugW]

Curiously that benchmark has the version running as iOS 17.6…
The other benchmark from the 10 core was definitely on 17.5

Good pickup.

For the 10-core models, it's 17.4 or 17.5. iPadOS 17.6 isn't even available as a beta, which makes me suspect the 9-core score is from Apple.

I’d like to see sustained scores. I find it hard to believe it will be able to do this consistently.

Since the ‘‘improved“ heat sink is now behind the Apple logo, hopefully someone can develop a stand with built-in an AIO water pump.

It wouldn't be able to sustain this performance indefinitely in an iPad Pro or even a MacBook Air, but I suspect it would be able to do it in a MacBook Pro.

 

[Lounge vibes 05]

Good pickup.

For the 10-core models, it's 17.4 or 17.5. iPadOS 17.6 isn't even available as a beta, which makes me suspect the 9-core score is from Apple.


It wouldn't be able to sustain this performance indefinitely in an iPad Pro or even a MacBook Air, but I suspect it would be able to do it in a MacBook Pro.

And certainly in the Mac mini.

 

[Confused-User]

at that time I’ll also have mine.. I’m just curious how to test it properly when arrives

My honest answer, not meant to be insulting, is that you shouldn't try. You don't begin to know enough to do any meaningful tests. This is a deep and subtle topic that has confounded a lot of really clever and knowledgeable people. Tackling it from a position of ignorance is not going to produce useful results.

You could download a copy of some commonly available test like GB6 and run it, but what's the point? Your scores will be roughly the same as all the other publically available scores.

I’d like to see sustained scores. I find it hard to believe it will be able to do this consistently.

Since the ‘‘improved“ heat sink is now behind the Apple logo, hopefully someone can develop a stand with built-in an AIO water pump.

You jest, but I was thinking about this the other day. Water-cooled is pointless, but a cooling stand might actually be a commercially viable product. I'm not sure though - the number of people needing top sustained performance on the iPP will be pretty small.

BTW, you can easily do a good job of this yourself by buying one of those metal meat defroster platters and running your ipad on that. It works really well. I used to do this back in the day with an old Intel MacBook when I was pushing it hard, and it had a dramatic effect on processor temps.

 

[ipaddaro]

My honest answer, not meant to be insulting, is that you shouldn't try. You don't begin to know enough to do any meaningful tests. This is a deep and subtle topic that has confounded a lot of really clever and knowledgeable people. Tackling it from a position of ignorance is not going to produce useful results.

You could download a copy of some commonly available test like GB6 and run it, but what's the point? Your scores will be roughly the same as all the other publically available scores.


You jest, but I was thinking about this the other day. Water-cooled is pointless, but a cooling stand might actually be a commercially viable product. I'm not sure though - the number of people needing top sustained performance on the iPP will be pretty small.

BTW, you can easily do a good job of this yourself by buying one of those metal meat defroster platters and running your ipad on that. It works really well. I used to do this back in the day with an old Intel MacBook when I was pushing it hard, and it had a dramatic effect on processor temps.

i think that a reliable test (and here’s the point of my question about what could be one), could help people decide what version to buy based on how they plan to use it

 

[jdb8167]

BTW, you can easily do a good job of this yourself by buying one of those metal meat defroster platters and running your ipad on that. It works really well. I used to do this back in the day with an old Intel MacBook when I was pushing it hard, and it had a dramatic effect on processor temps.

I’ve used a fan based laptop cooler with my M2 MacBook Air and it essentially prevented it from throttling. I’ve been meaning to test again with my M3 MBA and see if I get similar results.

 

[Wags]

That multi-core score is higher than I anticipated. In that case I will be sticking with my 256GB order. I will happily pocket the £600 difference.

Those scores would be almost double my M1.

 

[Confused-User]

i think that a reliable test (and here’s the point of my question about what could be one), could help people decide what version to buy based on how they plan to use it

We've already got those, to the extent that you consider things like GB6 reliable. Which they, within certain contexts and for certain purposes. And when embargo ends, and they ship, we'll have lots more.

 

[hovscorpion12]

These new Geekbench scores are wild. I can not believe these are coming from the M4 chip. On an iPad.

3934 on Single score and 14924 on Multi-Score. This could mean that the 53793 on the GPU Multi-Score could be higher, I cannot wait until Wednesday.

With scores like these, Final Cut, Davinci Resolve, Affinity Photo 2, LumaFusion should be flying through workloads.

 

[Melbourne Park]

....

BTW, you can easily do a good job of this yourself by buying one of those metal meat defroster platters and running your ipad on that. It works really well. I used to do this back in the day with an old Intel MacBook when I was pushing it hard, and it had a dramatic effect on processor temps.

Never heard of such a plate. I'd have thought a defroster would have a heater in it? If not, then what about a cheap alloy plate bought from the hardware store? Maybe sling a plastic bag over its back and fill it up with ice cubes ...

 

[EugW]

Regarding sustained performance, even a granite countertop helps, at least compared to a wood table.

This is how the 2017 MacBook Core m3, i5, i7 behave during sustained CPU load: 10+ runs of Cinebench

I posted this originally in a different thread, but I think it deserves its own thread. See this post for results of the m3, i5, and i7, with the m3 tested on two different table tops...

forums.macrumors.com

 

[klasma]

You jest, but I was thinking about this the other day. Water-cooled is pointless, but a cooling stand might actually be a commercially viable product. I'm not sure though - the number of people needing top sustained performance on the iPP will be pretty small.

This already exists: https://www.amazon.com/TACOMEGE-Aluminum-Writing-Foldable-Portable/dp/B09J8KVJ61/
Or even: https://www.amazon.com/zooparty-Magnetic-Compatible-Temperature-Streaming/dp/B0CHYY6WT6/

Of course, a heat-sink based solution would be preferable.

 

[theotherphil]

For sure…. But not sure how many people this would apply too.
Also I imagine the majority that do need this have a MacBook Pro for that workflow

What people forget is that the iPad Pro isn't aimed at the masses....it's aimed at the small percentage of people that need its performance, or are so cashed up to want its performance. Apple sell a number of other iPads with more than enough performance for the general population (the M2 iPad Air for example).

 

[Maven1975]

I would love good sustained performance. 🤓 Even if it leveled out at 90%! Unfortunately, I’m still using a 2018 12.9 IPP due to the screen not having PWM. Not holding my breath this year….

I’m still trying to figure out how stacking two OLED materials on top of each other is supposed to help with the issue of burn in and color saturation. 1000nits sustained and 1600nits peak is asking a lot. So is it design for the bottom material to be damaged in hopes that it protects the top material? Would we still not get uneven blotches visible from the lower surface?

 

[klasma]

I’m still trying to figure out how stacking two OLED materials on top of each other is supposed to help with the issue of burn in and color saturation. 1000nits sustained and 1600nits peak is asking a lot. So is it design for the bottom material to be damaged in hopes that it protects the top material? Would we still not get uneven blotches visible from the lower surface?

Presumably each layer only emits half the brightness, thus halving the wear. It's been said that tandem OLED quadruples the lifetime, so the effect is better than linear.

 

[throAU]

But the m1 is over kill for iPadOS, what does the app do to need that power, even a 2018 iPad Pro is fast enough

Games, running workloads faster so that the machine can go to sleep faster to save battery life.

Faster processing (at same/similar power draw) is always better to have, especially in a battery powered device.

 

[EugW]

I’m still trying to figure out how stacking two OLED materials on top of each other is supposed to help with the issue of burn in and color saturation. 1000nits sustained and 1600nits peak is asking a lot. So is it design for the bottom material to be damaged in hopes that it protects the top material? Would we still not get uneven blotches visible from the lower surface?

I'm not an engineer, but this is what I've learned.

1. The power utilization, heat generation, and risk of burn-in is not linearly related to brightness. From what I gather, the power utilization increases at a faster rate than the brightness, so as you go brighter the heat generation and risk of burn-in gets much, much higher. Consequently at any specific brightness, the power utilization and heat generation of two layers is actually less than a single layer at the same brightness. To put it another way, to generate say 500 nits, two layers emitting 250 nits will actually use less combined power and will generate less combined heat. So, the expected longevity of two layers is much more than twice as good than one layer at the same brightness.

2. Not all pixels will produce the same amount of light with the same amount of current. OLED can control the brightness of each pixel individually. With a single layer you may have to crank up the current to insensitive pixels to produce a desired amount of brightness. This puts it on the higher end of heat generation and risk of burn-in. With two layers you can balance the output to compensate. If one pixel is insensitive, you can still run it at a reasonable power level, knowing that it won't be be as bright as you normally would want it to be, if you also know the second layer has a more sensitive pixel. You can run the second pixel brighter to compensate, if that pixel doesn't have to have as much current pumped through it. That way the final brightness is still where you want it to be, but at lower power levels and lower heat generation.

 

[Confused-User]

Never heard of such a plate. I'd have thought a defroster would have a heater in it? If not, then what about a cheap alloy plate bought from the hardware store? Maybe sling a plastic bag over its back and fill it up with ice cubes ...

What I'm talking about it just a plate of metal. Basically a big heat sink. Turns out, that works well for equalizing cold temps with ambient as well as high temps, thus the defroster product. Your "cheap alloy plate" will work to the extent that it conducts well. And sure, the ice cubes will help. But if you need them, instead try one of the products @klasma suggested.

 

[Confused-User]

2. Not all pixels will produce the same amount of light with the same amount of current. OLED can control the brightness of each pixel individually. With a single layer you may have to crank up the current to insensitive pixels to produce a desired amount of brightness. This puts it on the higher end of heat generation and risk of burn-in. With two layers you can balance the output to compensate. If one pixel is insensitive, you can still run it at a reasonable power level, knowing that it won't be be as bright as you normally would want it to be, if you also know the second layer has a more sensitive pixel. You can run the second pixel brighter to compensate, if that pixel doesn't have to have as much current pumped through it. That way the final brightness is still where you want it to be, but at lower power levels and lower heat generation.

I'm a little skeptical about this (but also a lot ignorant, so...). What does the calibration? Also, this would lose (some of) the longevity benefit of two layers, for that pixel, and the working but relatively overloaded one would degrade faster than its neighbors, leaving it out of whack when it starts to degrade ahead of the rest.

 

[EugW]

I'm a little skeptical about this (but also a lot ignorant, so...). What does the calibration? Also, this would lose (some of) the longevity benefit of two layers, for that pixel, and the working but relatively overloaded one would degrade faster than its neighbors, leaving it out of whack when it starts to degrade ahead of the rest.

I’m not sure of the actual ratios or whatever but let’s say we rate power 1 to 10 and brightness 1 to 10 as well. Since power vs brightness isn’t linear, let’s say on average to reach brightness 5 you need power 3 but to reach brightness 10 you need to have power 10.

Single layer:
Pixel 1 - brightness 10, power 10
Really bad for pixel 1 longevity

Dual layer ideal with 2 equally sensitive pixels:
Pixel 1 - brightness 5, power 3
Pixel 2 - brightness 5, power 3
Very safe for pixel 1 & 2 longevity

Dual layer with one less sensitive pixel (not balanced for power):
Pixel 1 - brightness 5, power 6
Pixel 2 - brightness 5, power 3
Not good for Pixel 1 longevity
Very safe for Pixel 2 longevity

Dual layer with one less sensitive pixel (rebalanced for power):
Pixel 1 - brightness 4, power 4.5
Pixel 2 - brightness 6, power 4.5
Reasonably safe for both Pixel 1 & 2 longevity

These numbers are made up, but I hope it does illustrate how rebalancing between two pixels of different sensitivities can help.

 

[johannnn]

Does anyone know if a MBA/MBP with M4 will run hotter or cooler? Especially the MBA, I think the M2 was a bit hotter than the M1, and the M3 was a bit hotter than the M2. Which makes sense since it's a more powerful chip and no active cooling. But is this different with the M4? Or will the MBA soon be too hot if Apple creates more powerful chips every year?